I sat down at my computer yesterday morning with a hot cup of coffee, looking forward to the day’s work. Part of that work was to order a new Xserve for a video streaming project that I’m working on. You can imagine my surprise to discover Apple’s announcement:
Apple is transitioning away from Xserve. Xserve will be available for order through January 31, 2011… Apple offers two server solutions as an alternative to Xserve. Combine Mac OS X Server with a Mac Pro or Mac mini system…
As I read the 10-page document, and scoured the ‘net for the rest of the morning, desperately seeking more information about the development, my coffee went cold.
The enterprise IT community who are invested in Apple infrastructure has been going a little crazy since the announcement. Guessing Apple’s plans for future product development is a notoriously speculative venture, given their penchant for secrecy. It has always been a thorn in the side of enterprise IT planning. One has to try to read between the lines of flashy press conferences held by Steve Jobs, by market supply and demand numbers, patent filings, job postings, etc.. For example, this little video mashup has been making the rounds:
Most of that is from the D8 Conference last June. The question, started by Walt Mossberg at about 27:11, was about Apple’s business relationship with Google. The answer was that Apple is focussed on making good products, which may or may not include Google properties. Oh, and by the way, enterprise IT doesn’t always make good decisions as to what their users like/want/need. They’re “confused”.
Steve seems to be saying that IT managers don’t always make the best decisions about what technology is best for their users. So is Apple’s strategy to to put technology choice back in the hands of users by removing enterprise technology from the enterprise? Is it a coincidence that Apple’s massive new data centre is scheduled to open at approximately the same time that Xserves will no longer be sold?
“Cloud computing” is all the rage now in IT. Major organizations are moving their email and other services into the cloud. Why pay for soon-to-be-obsolete server hardware, the space, power, and cooling to run it, and the large salaries of system administrators to maintain it, when instead you can use the IT resources of Google, Microsoft, Amazon — or soon, Apple — for a nominal yearly fee?
Clouds are nice, but…
There is little doubt that Apple is about to expand it’s cloud computing offerings in a big way. However, not all services can be delegated to the cloud. There will be a strong need for in-house servers for the foreseeable future. To take just one example, organizations involved in video production and distribution require large amounts of high performance storage on site, and the computing power to process increasingly high-volume video data. The Internet just isn’t fast enough to handle that much data, yet.
A popular solution for those kinds of enterprise storage needs was Apple’s Xserve RAID and it’s companion software, XSan. Then one day, Apple announced it was also EOL. No more XRaid. A panic broke out — not as big as the one we’re seeing over the Xserve — but it was soon quelled by the availability of compatible third party products, such as the Promise RAID servers that Apple now sells on it’s website.
Essentially, Apple decided that other companies have better expertise in storage than they do, and so they exited the market. Other companies do a better job of it. Apple does not have expertise in search technology, so they incorporate Google or Bing into their products. As far as rackmount servers go, Apple’s Xserves aren’t the best in the class either.
Coming from a background working with Sun servers, the Xserve always seemed a little amateurish to me. They look nice, and the 1U form factor is great, but the onboard lights-out-management doesn’t even begin to compare with what the established server vendors offer. The 2 CPU and 3 disk limit is also underwhelming in comparison to other options. So maybe Apple will do the same thing they did with XRaid – let someone else do it better.
An easy way for Apple to make this transition would be to make another modification to their license agreement so that it not only allows for the virtualization of Mac OS X Server, but also allows it on non-Apple hardware. Even if they restricted virtualization to a specific brand or model of server, it would not only quell the current panic, but it would also make OS X Server a more attractive option in the increasingly virtualized enterprise IT environment.
Whither OS X Server 10.7?
As Steven Sande pointed out over on TUAW, when the upcoming Mac OS X 10.7 (“Lion”) was announced, there was no mention at all of the server version of the OS. That is unusual. I seem to recall Apple including both versions of the OS in prior release announcements.
Sande speculated that maybe OS X Server will be broken down into it’s component services, each of which will be offered up in the upcoming Mac App Store as add-ons for OS X 10.7. So a small business could turn a Mac Pro or Mini into a server simply by purchasing the module for the service that they need.
That’s a clever idea, but I don’t think that’s what’s going on. Apple has invested a lot into the development of OS X Server as a distinct product, and more importantly, the main users of that product represent core markets for Apple: the creative/media industry, and the education industry. They need reliable server hardware, and lots of it.
Furthermore, all of those Mac desktops in university labs and production houses need to be managed, and OS X Server is presently the only way to do that. Sure, Mac clients can login to Microsoft’s Active Directory, but managing preferences, home directories, etc., still requires OS X Server. It’s not going away, even if it must be run on a desktop computer instead of a proper server. If it does have to be run on a desktop computer though, the extra risk of downtime and visits to the server room makes it harder to justify the investment in Apple, iMacs included.
Worse case scenario
The worse case scenario is that the business press is right, and Apple wants out of the enterprise market altogether. If this is so, organizations with Mac server deployments will likely be looking at migrating to other platforms. In fact, they already are.
The Mac Pro is simply not a replacement for the Xserve. Those who invested in Xserve did so because it provided reliability: two power supplies, hot-swappable hard drives, and remote lights-out-management. Not to mention that it only takes up 1U of rack space and uses low power. Using a Mac Pro instead is likely too risky for many of these deployments, which in turn calls into question the use of Macs in the first place.